Which is more effective: running full-layer tests with an in-memory database or using integration and manual tests? Should developers complete unit, integration, and end-to-end tests before submitting a PR?
hey guys, i ve found that using real integreation tests gives a more reliable picture of what might go wrong. unit tests are fast but sometimes skip the full story. i like keeping a little bit of both, not over testing every time.
hey, i reckon combining methods works. in-mem tests speed things up, yet integration tests show real world quirks. ever noticed any trade-offs between speed n depth? would love to c how you all juggle this when prioritizing PR submissions.
Based on my experience, using an in-memory database for unit tests provides speedy feedback and helps quickly pinpoint issues in data handling. However, it lacks the depth of interaction typically present in full-layer integration tests. Thorough testing before PR submission is essential; unit tests verify individual components, integration tests confirm that system interactions function as expected, and manual checks catch anomalies that even automated tests can overlook. A balanced testing approach increases confidence in the codebase, ultimately reducing unexpected production issues.
hey ppl, i think a mix works best. in-mem tests run fast while integration tests help catch hidden bugs. some manuul checks still add value. how do u deal with false positives, and what’s your fav method? lets chat.